hvbench is an open and scalable OpenFlow controller/switch/hyperisor benchmark based on distributed message generators. A central API configures the message generators (which act as switches or controllers) and collects run-time statistics such as the response delay. Check out the OSSN paper here for an in-depth overview.

Following repositories are related to this project:

• hvbench-api: The Python API of hvbench
• hvmonitor: Monitor the resource comsumption of a network hypervisor
• tutorial: Tutorial on how to set-up and operate hvbench

• Distributed operation for scalability
• Centrally controlled & configured through etcd
• Add, reconfigure and remove tenants at run-time
• Add, connect, remove switches at run-time (work in progress)
• Static or exponential (Poisson) inter-arrival times
• One message type or mix of message types based on weights
• Support for OpenFlow 1.0 (1.3 in working)

See hvbench-tutorial for tutorials on how to set-up and operate hvbench.

See QUICKSTART.md. Setting up and operating hvbench is rather complex, we recommend to follow the tutorials first.

See COMPILE.md for instructions how to compile hvbench and how to build the docker image.

If you are using hvbench for scientific publications, please cite the following paper:

C. Sieber, A. Blenk, A. Basta and W. Kellerer, hvbench: An open and scalable SDN Network Hypervisor Benchmark, NetSoft 2016 - Workshop on OSSN

Bibtex:

@INPROCEEDINGS{sieber16hvbench,
    AUTHOR="Christian Sieber and Andreas Blenk and Arsany Basta and Wolfgang Kellerer",
    TITLE="hvbench: An open and scalable SDN Network Hypervisor Benchmark",
    BOOKTITLE="NetSoft 2016 - Workshop on OSSN () ",
    MONTH="jun",
    YEAR="2016",
}